AI analysis for the superficial
At the Adlershof Founder’s Lab the start-up ProMetronics is working on flawless functional microsurfaces
The young Adlershof company ProMetronics UG combines 3D microscopy with artificial intelligence to inspect the quality of functional microsurfaces already in the manufacturing process. If defects are detected, the AI findings serve as a basis for immediately readjusting the processes.
André Kempe and Ioannis Karakatsanis of ProMetronics UG are experts in deep analyses of surfaces. Using 3D microscopes and artificial intelligence, the two founders are exploring the question of how micro- and nanostructures influence the respective functions of surfaces – and how those functions can be targetedly optimised.
It may sound like basic research, but it has very practical relevance: Kempe was approached by a manufacturer of laboratory supplies. They had a problem with their “96-well microplates”, namely the plastic plates with 96 cavities typically used for high throughput screening in biotech and pharmaceutical labs. In nearly 20 percent of microplates, the tests would run poorly or not at all. There was no discernible pattern or cause for the error.
Kempe has been running a laboratory for material characterisation for many years. He is specialised in the analysis of functional surfaces. “We use physical and optical 3D microscopy at resolutions in the sub-nanometre range,” he explains. He also put the faulty microplates through corresponding analyses. Given their unit prices between one and 25 euros, it was immediately clear that the problem required a highly efficient testing method. “This is where Ioannis came in as an AI expert,” Kempe reports.
At the nanometre scale, each of the 96 wells in the plastic plate looks like a mountainscape with unique structures. The co-founders translated these into mathematical vector models – and also recorded the quality of the surface function in a number-based evaluation scheme. That allowed them to mathematically describe optimal properties of a well, and to automatically compare other wells with this ideal. Within ten seconds, the 3D microscope scans the structure and AI algorithms complete the comparison. “This test procedure can be used in the manufacturing process for tight-knit random checks,” Kempe explains.
That is not all the AI can do. It also establishes the relationship between its complex structural analysis and evaluation and the respective process parameters. Be it the temperature and pressure of injection moulding, molecular composition of polymers or downstream microstructuring and coating processes.
Even if the part costs are low in the case of the microplates, the potential leverage is huge. If 20 out of 100 plates are randomly defective, then this adds unnecessary uncertainty into the in-vitro cultivation of stem cells, the development of vaccines and tests that determine whether substances repel or bind proteins. Manufacturers can even make the previously rigid purchasing process more flexible if AI quickly guides them towards optimal process parameters. “And, of course, our solution can also ensure the quality of other functional surfaces and their manufacturing processes,” says Kempe.
The Adlershof Founder’s Lab is supporting ProMetronics with a grant. Development is coming along well. The team is now looking for a business angel with good industry contacts. After all, despite all the AI, they know there is one thing that matters most when they start marketing: the human factor.
By Peter Trechow for Adlershof Journal